Connection-Oriented Networks - Harry Perros 1
Chapter 2:SONET/SDH and GFP
TOPICS– T1/E1– SONET/SDH - STS 1, STS -3 frames– SONET devices– Self-healing rings– Generic frame protocol, and Data over SONET
Connection-Oriented Networks - Harry Perros 2
T1/E1
• Time division multiplexing allows a linkto be utilized simultaneously by manyusers
MUX
DEMUX
N input links
N output links
link
12
N
12
N
Connection-Oriented Networks - Harry Perros 3
• The transmission is organized into frames.• Each frame contains a fixed number of time slots.• Each time slot is pre-assigned to a specific input
link. The duration of a time slot is either a bit or abyte.
• If the buffer of an input link has no data, then itsassociated time slot is transmitted empty.
• A time slot dedicated to an input link repeatscontinuously frame after frame, thus forming achannel or a trunk.
Connection-Oriented Networks - Harry Perros 4
Pulse code modulation
• TDM is used in telephony• Voice analog signals are digitized at the
end office using Pulse Code Modulation.• A voice signal is sampled 8000 times/sec,
or every 125 µsec.• A 7-bit or 8-bit number is created every
125 µsec.
Connection-Oriented Networks - Harry Perros 5
The Digital Signal (DS) andITU-T standard
• A North American standard that specifies how tomultiplex several voice calls onto a single link.
• The DS standard is a North American standard andit is not the same as the international hierarchystandardized by ITU-T.
• Both standards are independent of the transmission.
Connection-Oriented Networks - Harry Perros 6
T carrier / E carrier• The DS signal is carried over a carrier system
known as the T carrier.– T1 carries the DS1 signal,– T2 carries the DS2 signal etc
• The ITU-T signal is carried over a carrier systemknown as the E carrier.
• The DS and ITU-T hierarchy is known as theplesiochronous digital hierarchy (PDH). (Plesionmeans “nearly the same”, and chronos means“time” in Greek).
Connection-Oriented Networks - Harry Perros 7
Digital signal number Voice channels Data Rate (Mbps)DS0 1 0.064DS1 24 1.544
DS1C 48 3.152DS2 96 6.312DS3 672 44.736
DS3C 1344 91.053DS4 4032 274.176
Table 2.1: The North American Hierarchy
Level number Voice channels Data Rate (Mbps)0 1 0.0641 30 2.0482 120 8.4483 480 34.3684 1920 139.2645 7680 565.148
Table 2.2: The international (ITU-T) hierarchy
Connection-Oriented Networks - Harry Perros 8
The DS1 signal
• 24 8-bit time slots/frame– Each time slot carries 8 bits/ 125 µsec, or the channel
carries a 64 Kbps voice.– Every 6th successive time slot (i.e, 6th, 12th, 18th,
24th, etc), the 8 bit is robbed and it is used forsignaling.
• F bit: Used for synchronization. It transmits thepattern: 10101010…
FTime
slot 1
Time
slot 2
Time
slot 3
Time
slot 24. . .
Connection-Oriented Networks - Harry Perros 9
• T1:– Total transmission rate: 24x8+1 = 193 bits per 125 µ
sec, or 1.544 Mbps• E1
– 30 voice time slots plus 2 time slots forsynchronization and control
– Total transmission rate: 32x8 = 256 bits per 125 µsec,or 2.048 Mbps
Connection-Oriented Networks - Harry Perros 10
Fractional T1/E1
• Fractional T1 or E1 allows the use of onlya fraction of the T1 or E1 capacity.
• For example: if N=2, then only two timeslots are used per frame, which correspondsto a channel with total bandwidth of 128Kbps.
Connection-Oriented Networks - Harry Perros 11
Unchannelized frame signal
• The time slot boundaries are ignored by thesending and receiving equipment.
• All 192 bits are used to transport data followed bythe 193rd framing bit.
• This approach permits more flexibility intransmitting at different rates.
• This scheme is implemented using proprietarysolutions.
Connection-Oriented Networks - Harry Perros 12
The synchronous optical network(SONET)
• Proposed by Bellcore (Telecordia).– It was designed to multiplex DS-n signals and
transmit them optically.• ITU-T adopted the synchronous digital
hierarchy (SDH), as the internationalstandard.– It enables the multiplexing of level 3 signals
(34.368 Mbps)
Connection-Oriented Networks - Harry Perros 13
STS, STM, OC
• The electrical side of the SONET signal isknown as the synchronous transport signal(STS)
• The electrical side of the SDH is known asthe synchronous transport module (STM).
• The optical side of a SONET/SDH signal isknown as the optical carrier (OC).
Connection-Oriented Networks - Harry Perros 14
The SONET/SDH hierarchyOptical
level
SONET
level
(electrical)
SDH
level
(electrical)
Data rate
(Mbps)
Overhead
rate
(Mbps)
Payload
rate
(Mbps)
OC-1 STS-1 - 51.840 1.728 50.112
OC-3 STS-3 STM-1 155.520 5.184 150.336
OC-9 STS-9 STM-3 466.560 15.552 451.008
OC-12 STS-12 STM-4 622.080 20.736 601.344
OC-18 STS-18 STM-6 933.120 31.104 902.016
OC-24 STS-24 STM-8 1244.160 41.472 1202.688
Oc-36 STS-36 STM-12 1866.240 62.208 1804.932
OC-48 STS-48 STM-16 2488.320 82.944 2405.376
OC-96 STS-96 STM-32 4976.640 165.888 4810.752
OC-192 STS-192 STM-64 9953.280 331.776 9621.504
OC-768 STS-768 STM-256 39813.120 1327.104 38486.016
OC-N STS-N STM-N/3 N*51.840 N*1.728 N*50.112
Connection-Oriented Networks - Harry Perros 15
• SONET/SDH is channelized.– STS-3 consists of 3 STS-1 streams, and each STS-
1 consists of a number of DS-1 and E1signals.– STS-12 consists of 12 STS-1 streams
• Concatenated structures (OC-3c, OC-12c, etc)– The frame of the STS-3 payload is filled with
ATM cells or IP packets packed in PPP or HDLCframes.
– Concatenated SONET/SDH links are commonlyused to interconnect ATM switches and IP routers(Packets over SONET).
Connection-Oriented Networks - Harry Perros 16
The STS-1 frame structure1 2 3 4 5 6 … 90
1 1 2 3 4 5 6 … 90
2 91 92 93 94 95 96 … 180
3 181 182 183 184 185 186 … 270
4 271 272 273 274 275 276 … 360
5 361 362 363 364 365 366 … 450
6 451 452 453 454 455 456 … 560
7 561 562 563 564 565 566 … 630
8 631 632 636 634 635 636 … 720
9 721 722 723 724 725 726 … 810
Connection-Oriented Networks - Harry Perros 17
• Main features– The frame is presented in matrix form and it is
transmitted row by row.– Each cell in the matrix corresponds to a byte– The first three columns contain overheads– The remaining 87 columns carry the
synchronous payload envelope (SPE), whichconsists of user data, and additional overheadsreferred to as the payload overhead (POH)
Connection-Oriented Networks - Harry Perros 18
An SPE may straddle betweentwo successive frames
Frame i
Frame i+1
1 2 3 4 5 6 . . . 901
2
3
4
5
6
7
8
9
276
276275
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
1
2
3
4
5
6
7
8
9
Connection-Oriented Networks - Harry Perros 19
The section, line, and path overheads
Section
Line
STS-1 STS-1
A B
regeneratorregeneratorSTS-1
A1
A12
STS-12
. . .
STS-1
B1
B12STS-12
. . .
Section Section Section Section
LineLine
Path
Connection-Oriented Networks - Harry Perros 20
• Section: a single link with a SONET deviceor a regenerator on either side of it.
• Line: A link between two SONET devices,which may include regenerators
• The section overhead in the SONET frameis associated with the transport of STS-1frames over a section, and the lineoverhead is associated with the transport ofSPEs over a line.
Connection-Oriented Networks - Harry Perros 21
The SONET stack
Section
Line
Path
Photonic
Section
Line
Path
Photonic
Section
Line
Photonic
Section
Photonic
Section
Photonic
Section
Line
Photonic
AiA Regenerator Regenerator BiB
Connection-Oriented Networks - Harry Perros 22
STS-1: Section and line overheads
SOH
LOH
Column1 2 3
1 A1 A2 J02 B1 E1 F13 D1 D2 D34 H1 H2 H35 B2 K1 K26 D4 D5 D67 D7 D8 D98 D10 D11 D129 Z1 Z2 E2
Connection-Oriented Networks - Harry Perros 23
• The following are some of the bytes in thesection overhead (SOH) :– A1 and A2: These two bytes are called the
framing bytes and they are used for framealignment. They are populated with the value1111 0110 0010 1000 or 0xF628, whichuniquely identifies the beginning of an STS-frame.
– J0: This is called the section trace byte and itis used for to trace the STS-1 frame back to itsoriginating equipment.
Connection-Oriented Networks - Harry Perros 24
– B1: This byte is the bit interleaved parity byteand it is commonly referred to as BIP-8. It isused to perform an even-parity check on theprevious STS-1 frame after the frame has beenscrambled. The parity is inserted in the BIP-8field of the current frame before it is scrambled
– E1: This byte provides a 64 Kbps channel canbe used for voice communications by fieldengineers.
Connection-Oriented Networks - Harry Perros 25
• The following are some of the bytes in the lineoverhead (LOH) that have been defined:– H1 and H2: These two bytes are known as the pointer
bytes, and they contain a pointer that points to thebeginning of the SPE within the STS-1 frame. Thepointer gives the offset in bytes between the H1 andH2 bytes and the beginning of the SPE.
– B2: This is similar to the B1 byte in the sectionoverhead and it is used to carry the BIP-8 parity checkperformed on the line overhead section and thepayload section. That is, it is performed on the entireSTS-1 frame except the section overhead bytes.
Connection-Oriented Networks - Harry Perros 26
The path overhead bytes
J1
B3C2G1F2
H4Z3
Z4Z5
J1B3C2G1F2H4 Z3Z4Z5
Location of the POH The POH bytes
Connection-Oriented Networks - Harry Perros 27
• The following are some of the bytes that havebeen defined:– B3: This byte is similar to B1 used in the section
overhead and B2 used in the line overhead. It is used tocarry the BIP-8 parity check performed on the payloadsection. That is, it is performed on the entire STS-1frame except the section and line overhead bytes.
– C2: This byte is known as the path signal label and itindicates the type of user information carried in theSPE, such as, virtual tributaries (VT), asynchronousDS-3, ATM cells, HDLC-over-SONET, and PPP overSONET.
Connection-Oriented Networks - Harry Perros 28
The STS-1 payload
• The payload consists of user data and thepath overhead.
• User data:– Virtual tributaries: sub-rate synchronous data
streams, such as DS-0, DS-1, E1, and entireDS-3 frames
– ATM cells and IP packets
Connection-Oriented Networks - Harry Perros 29
Virtual tributaries
• The STS-1 payload is divided into sevenvirtual tributary groups (VTG).
• Each VTG consists of 108 bytes (12 columns)• Each VTG may carry a number of virtual
tributaries, i.e., sub-rate streams.
Connection-Oriented Networks - Harry Perros 30
• The following virtual tributaries have beendefined:– VT1.5: This virtual tributary carries one DS-1
signal and it is contained in three columns, thattake up 27 bytes. Four VT1.5’s can betransported in a single VTG.
– VT2: This virtual tributary carries an E1 signalof 2.048 Mbps. VT2 is contained in fourcolumns, that is it takes up 36 bytes. ThreeVT2’s can be carried in a single VTG.
Connection-Oriented Networks - Harry Perros 31
• VT3: This virtual tributary transports theunchannelized DS-1 signal. A VT3 iscontained in 6 columns that takes up 54 bytes.This means that a VTG can carry two VT3s.
• VT6: This virtual tributary transports a DS-2signal, which carries 96 voice channels. VT6 iscontained in 12 columns, that is it takes up 108bytes. A VTG can carry exactly one VT2.
Connection-Oriented Networks - Harry Perros 32
ATM cells
• Mapped directly onto the SPE. An ATMcells may straddle two SPEs.
10Cell 1 Cell 2
Cell 2 Cell 3
Cell 14 Cell 15
Cell 15
9041
9
2
8
3
POH
Connection-Oriented Networks - Harry Perros 33
IP packet over SONET
• IP packets are first encapsulated in HDLC andthe resulting frames are mapped into the SPEpayload row by row as in the case above forATM cels.
10 9041
9
2
8
3POH
7E 7E 7E
7E7E7E
Connection-Oriented Networks - Harry Perros 34
• IP packets can also be encapsulated in PPPinstead of HDLC.
• A frame may straddle over two adjacent SPEs, asin the case of ATM.
• The interframe fill 7E is used to maintain acontinuous bit tstream
Connection-Oriented Networks - Harry Perros 35
The STS-3 frame structure
Overhead section Payload section
1 2 3 4 5 6 7 8 9 10 11 12270
. . .
1st S
TS-1
1st S
TS-1
1st S
TS-1
1st S
TS-1
1st S
TS-1
2nd S
TS-1
2nd S
TS-1
2nd S
TS-1
2nd S
TS-1
2nd S
TS-1
3rd S
TS-1
3rd S
TS-1
3rd S
TS-1
3rd S
TS-1
3rd S
TS-1
Connection-Oriented Networks - Harry Perros 36
• The channelized STS-3 frame is constructed bymultiplexing byte-wise three channelized STS-1frames. As a result:– Byte 1, 4, 7, … , 268 of the STS-3 frame contains byte
1, 2, 3, … , 90 of the first STS-1 frame.– Byte 2, 5, 8, …, 269 of the STS-3 frame contains byte
1, 2, 3, … , 90 of the second STS-1 frame– Byte 3, 6, 9, …, 270 of the STS-3 frame contains byte
1, 2, 3, … , 90 of the third STS-1 frame.• This byte-wise multiplexing, causes the columns
of the three STS-1 frames to be interleaved in theSTS-3 frame
Connection-Oriented Networks - Harry Perros 37
• The first 9 columns of the STS-3 framecontain the overhead part and theremaining columns contain the payloadpart.
• Error checking and some overhead bytesare for the entire STS-3 frame, and they areonly meaningful in the overhead bytes ofthe first STS-1 frame.
Connection-Oriented Networks - Harry Perros 38
SONET/SDH devices
• Several different equipment exist:– Terminal multiplexer (TM)– Add/drop multiplexer (ADM)– Digital cross connect (DCS)
Connection-Oriented Networks - Harry Perros 39
• It multiplexes a number of DS-n or E1 signalsinto a single OC-N signal
• It consists of a controller, low-speed interfacesfor DS-n or E1 signals, an OC-N interface, and atime slot interchanger (TSI)
• It works also as a demultiplexer
. . .
DS-n
OC-N
DS-n
TM
The terminal multiplexer (TM):
Connection-Oriented Networks - Harry Perros 40
• It is a more complex version of the TM• It receives an OC-N signal from which it can
demultiplex and terminate (i.e., drop) anynumber of DS-n or OC-M signals, where M<N,while at the same time it can add new DS-n andOC-M signals into the OC-N signal.
. . .
DS-n. OC-M
OC-N OC-NADM
The add/drop multiplexer (ADM)
Connection-Oriented Networks - Harry Perros 41
SONET ringsADM
1ADM
2
ADM3
ADM4
OC3
OC3
OC3
OC3
• SONET/SDH ADM devices are typically connected toform a SONET/SDH ring.
• SONET/SDH rings are self-healing, that is they canautomatically recover from link failures.
Connection-Oriented Networks - Harry Perros 42
An example of a connection
A
B
TM1
TM2
ADM1
ADM2
ADM3
ADM4
DS1
OC12
DS1
OC12
OC12
OC12
OC3
OC3
Connection-Oriented Networks - Harry Perros 43
• A transmits a DS-1 signal to TM 1• TM 1 transmits an OC-3 signal to ADM 1• ADM 1 adds the OC-3 signal into the STS-
12 payload and transmits it out to the nextADM.
• At ADM 3, the DS-1 signal belonging to Ais dropped from the payload andtransmitted with other signals to TM 2.
• TM 2 in turn, demultiplexes the signals andtransmits A’s DS-1 signal to B.
Connection-Oriented Networks - Harry Perros 44
• Connection setup:– Using network management procedures the
SONET network is provisioned appropriately.This is an example of a permanent connection.
– It remains up for a long time.• The connection is dedicated to user A
whether the user transmits or not.
Connection-Oriented Networks - Harry Perros 45
A digital cross connect (DCS)
Ring 1 Ring 2ADM
ADM
ADM
ADM
ADM
ADM
DCS
• It is used to interconnect multiple SONET rings• It is connected to multiple incoming and outgoing OC-N
interfaces. It can drop and add any number of DSn and/orOC-M signals, and it can switch DSn and/or OC-Msignals from an incoming interface to any outgoing one.
Connection-Oriented Networks - Harry Perros 46
Self-healing SONET/SDH rings
• SONET/SDH rings have been speciallyarchitected so that they are available 99.999% ofthe time (6 minutes per year!)
• Causes for ring failures:– Fiber link failure due to accidental cuts, and
transmitter/receiver failure– SONET/SDH device failure (rare)
Connection-Oriented Networks - Harry Perros 47
Automatic protection switching (APS)
• SONET/SDH rings are self-healing, that is, thering’s services can be automatically restoredfollowing a link failure or degradation in thenetwork signal.
• This is done using the automatic protectionswitching (APS) protocol. The time to restore theservices has to be less than 50 msec.
Connection-Oriented Networks - Harry Perros 48
Protection schemes: point-to-point
• Schemes for link protection– dedicated 1+1– 1:1– Shared 1:N
ADM
Working
ProtectionADM
Connection-Oriented Networks - Harry Perros 49
Working/protection fibers
• The working and protection fibers have tobe diversely routed. That is, the two fibersuse separate conduits and different physicalroutes.
• Often, for economic reasons, the two fibersuse different conduits, but they use thesame physical path. In this case, we saythat they are structurally diverse.
Connection-Oriented Networks - Harry Perros 50
Classification of self-healing rings
• Various ring architectures have beendeveloped based on the following threefeatures:– Number of fibers
• 2 or 4 fibers– Direction of transmission:
• Unidirectional bidirectional– Line or path switching
Connection-Oriented Networks - Harry Perros 51
Number of fibers: 2- or 4-fiber rings
Two-fiber ring: fibers 1, 2, 3, and 4 areused to form the working ring (clockwise),and fibers 5, 6, 7, and 8 are used to formthe protection ring (counter-clockwise).
1
2
3
4
5
6
7
8
ADM 1 ADM 2
ADM 3ADM 4
ADM 1 ADM 2
ADM 3ADM 4
Connection-Oriented Networks - Harry Perros 52
• In another variation of the two-fiber ring, each set of fibersform a ring which can be both a working and a protectionring. In this case, the capacity of each fiber is divided intotwo equal parts, one for working traffic and the other forprotection traffic.
• In a four-fiber SONET/SDH ring there are two workingrings and two protection rings, one per working ring.
1
2
3
4
5
6
7
8
ADM 1 ADM 2
ADM 3ADM 4
ADM 1 ADM 2
ADM 3ADM 4
Connection-Oriented Networks - Harry Perros 53
Direction of transmission
• Unidirectional ring:– signals are only transmitted in one
direction of the ring.• Bidirectional ring:
– signals are transmitted in both directions.
Connection-Oriented Networks - Harry Perros 54
Line and path switching
• Path switching: Restores the traffic on thepaths affected by a link failure (a path is anend-to-end connection between the pointwhere the SPE originates and the point whereit terminates.)
• Line switching: Restores all the traffic thatpasses through a failed link.
Connection-Oriented Networks - Harry Perros 55
Based on these three features, we have thefollowing 2-fiber or 4-fiber possible ringarchitectures:– Unidirectional Line Switched Ring (ULSR)– Bidirectional Line Switched Ring (BLSR)– Unidirectional Path Switched Ring (UPSR)– Bidirectional Path Switched Ring (BPSR)
Connection-Oriented Networks - Harry Perros 56
Of these rings the following three areused:– Two-fiber unidirectional path switched ring
(2F-UPSR)– Two-fiber bidirectional line switched ring
(2F-BLSR)– Four-fiber bidirectional line switched ring
(4F-BLSR)
Connection-Oriented Networks - Harry Perros 57
Two-fiber unidirectionalpath switched ring (2F-UPSR)
ADM 1 ADM 2
ADM 3ADM 4
5
264 8
3
7
A
Protection ring
Working ring
1B
Connection-Oriented Networks - Harry Perros 58
• Features:– Working ring consists of fibers 1, 2, 3 and 4,
and the protection ring of fibers 5, 6, 7, and 8.– Unidirectional transmission means that traffic
is transmitted in the same direction. Atransmits to B over fiber 1 of the working ring,and B transmits over fibers 2, 3, and 4 of theworking ring.
– Used as a metro edge ring to interconnectPBXs and access networks to a metro core ring
Connection-Oriented Networks - Harry Perros 59
• Self-healing mechanism:– Path level protection using the 1+1 scheme. The
signal transmitted by A is split into two. Onecopy is transmitted over the working fiber 1, andthe other copy is transmitted over the protectionfibers 8, 7, and 6.
– During normal operation, B receives twoidentical signals from A, and selects the onewith the best quality. If fiber 1 fails, B willcontinue to receive A’s signal over theprotection path. The same applies if there is anode failure.
Connection-Oriented Networks - Harry Perros 60
Two-fiber bidirectional line switchedring (2F-BLSR)
ADM 1 ADM 2 ADM 3
ADM 4
7
396 12
5
11
A B1
8
4
2
10
ADM 5ADM 6
C
Connection-Oriented Networks - Harry Perros 61
• Features:– Used in metro core rings.– Fibers 1, 2, 3, 4, 5, and 6 form a ring, call it ring 1, on
which transmission is clockwise. Fibers 7, 8, 9, 10, 11,and 12 form another ring, call it ring 2, on whichtransmission is counter-clockwise.
– Both rings 1 and 2 carry working and protection traffic.This is done by dividing the capacity of each fiber onring 1 and 2 to two parts. One part is used to carryworking traffic and the other protection traffic.
– A transmits to B over the working part of fibers 1 and2 of ring 1, and B transmits to A over the working partof fibers 8 and 7 of ring 2.
Connection-Oriented Networks - Harry Perros 62
• Self-healing mechanism:– The ring provides line switching. If fiber 2 fails
then the traffic that goes over fiber 2 will beautomatically switched to the protection part ofring 2.
– That is, all the traffic will be re-routed to ADM3 over the protection part of ring 2 using fibers7, 12, 11, 10, and 9. From there, the traffic foreach connection will continue on following theoriginal path of the connection.
Connection-Oriented Networks - Harry Perros 63
Four-fiber bidirectional line switchedring (4F-BLSR)
Working rings
ADM 1 ADM 2 ADM 3
ADM 4
A B
ADM 5ADM 6
Protection rings
Connection-Oriented Networks - Harry Perros 64
• Features– Two working rings and two protection rings.
The two working rings transmit in oppositedirections, and each is protected by aprotection ring which transmits in the samedirection.
– The advantage of this four-fiber ring is that itcan suffer multiple failures and still function.In view of this, it is deployed by long-distancetelephone companies in regional and nationalrings.
Connection-Oriented Networks - Harry Perros 65
• Self-healing operation (span switching):– If a working fiber fails, the working traffic will
be transferred over its protection ring. This isknown as span switching.
ADM 1 ADM 2 ADM 3 ADM 1 ADM 2 ADM 3
Normal operation Span switching
Connection-Oriented Networks - Harry Perros 66
• Self-healing operation (ring switching):– Often, the working and protection fibers are
part of the same bundle of fibers. When thebundle is cut the traffic will be switched to theprotection fibers. This is known as ringswitching.
B
ADM 1 ADM 2 ADM 3
ADM 4
A
ADM 5ADM 6
Working
Protection
ADM 1 ADM 2 ADM 3
ADM 4
A
B
ADM 5ADM 6
Working
Protection
B
Connection-Oriented Networks - Harry Perros 67
Generic Framing Procedure (GFP)
• This is a light-weight adaptation schemethat permits the transmission of differenttypes of traffic over SONET/SDH and inthe future, over G.709.
Connection-Oriented Networks - Harry Perros 68
• GFP permits the transport ofa) frame-oriented traffic, such as Ethernet, andb) block-coded data for delay-sensitive storage
area networks (SAN) transported by networkssuch as Fiber Channel, FICON, and ESCON
over SONET/SDH and G.709.• GFP is a result of joint standardization
effort by ANSI committee T1X1.5 and ITU-T.
• It is described in ITU-T recommendationG.7041
Connection-Oriented Networks - Harry Perros 69
Privatelines Ethernet ESCON FICON Fiber
Channel
Frame Relay POS
ATM
SONET/SDH
WDM/OTN
GFP
Voice Data (IP, MPLS, IPX) SAN
DM
Video
Existing and GFP-based transport options for end-user applications
HDLC
Connection-Oriented Networks - Harry Perros 70
The GFP stack
GFP
GFP client-dependent aspects
GFP client-independent aspects
SONET/SDH G.709
Ethernet IP over PPP SAN data
Connection-Oriented Networks - Harry Perros 71
GFP frame structure
Payload
Core header
Payload lengthPayload length
Core HECCore HEC
Payload header
Payload
Payload FCS
• GFP core header– Payload length indicator
(PLI) - 2 bytes. It gives thesize of the payload.
– Core HEC (cHEC) - 2bytes. It protects the PLIfield. Standard CRC-16enables single bit errorcorrection and multiple biterror detection.
Connection-Oriented Networks - Harry Perros 72
The GFP payload structure
Payload header
Payload
Payload FCS
Payload type
Payload type
Type HEC
Type HEC
0-60 bytesof
extension header
Payload FCS
Payload FCS
Payload FCS
Payload FCS
PTI
UPI
PFI EXI
Connection-Oriented Networks - Harry Perros 73
GFP payload headervariable-length area from 4 to 64 bytes.
• Payload type - 2 bytes– Payload type identifier (PTI) - 3 bits.
Identifies the type of frame:• User data frames , Client mgmt frames
– Payload FCS indicator (PFI) - 1 bit.Identifies if there is a payload FCS
– Extension header identifier (EXI) - 4 bits.Identifies the type of extension header.
– User payload identifier (UPI) - 8 bits.Identifies the type of payload
• Frame-mapped Ethernet• Frame-mapped PPP (IP, MPLS)• Transparent-mapped Fiber Channel• Transparent-mapped FICON• Transparent-mapped ESCON• Transparent-mapped GbE
• Type HEC (tHEC) - 2 bytes. It protects thepayload header. Standard CRC-16.
Payload type
Payload type
Type HEC
Type HEC
0-60 bytesOf
Extension header
PTI
UPI
PFI EXI
Connection-Oriented Networks - Harry Perros 74
GFP payload trailer
Payload header
Payload
Payload FCS
Payload FCS
Payload FCS
Payload FCS
Payload FCS
• Optional 4-byte FCS.– CRC-32– Protects the contents of
the payloadinformation field.
Connection-Oriented Networks - Harry Perros 75
GFP-client independent functions
• The client independent sublayer supportsthe following functions:– Frame delineation– Client/frame multiplexing– Payload scrambler– Client managment
Connection-Oriented Networks - Harry Perros 76
Frame delineation
• The framedelineationmechanism is similarto the one used inATM.
• The cHEC is used toassure correct frameboundaryidentification
hunt
Presync
Sync
CorrectcHEC
2ndcHEC match
Non-correctablecore header error
No 2ndcHEC
Connection-Oriented Networks - Harry Perros 77
• Operation:– Under normal conditions, the GFP receiver
operates in the Sync state. The receiverexamines the PLI field, validates the cHEC,and extracts the framed higher-level PD. Itthen moves on to the next GFP header.
– When an uncorrectable error in the coreheader occurs (i.e., cHEC fails and more thanone bit error is detected), the receiver entersthe Hunt state.
Connection-Oriented Networks - Harry Perros 78
• Hunt state:– Using the cHEC it attempts to locate the
beginning of the next GFP PDU, moving onebit at a time (Same as in ATM - see Perros “Anintroduction to ATM networks, Wiley 2001.
– Once this is achieved it moves to the Pre-Syncstate, where it verifies the beginning of theboundary of the next N GFP PDUs.
– If successful, it moves to the Sync state,otherwise it moves back to the hunt state.
Connection-Oriented Networks - Harry Perros 79
Frame multiplexing
• Client data frames and client managementframes are multiplexed, with client dataframes having priority over clientmanagement frames.
• Idle frames are inserted to maintain acontinuous bit flow (rate coupling)
Connection-Oriented Networks - Harry Perros 80
GFP client-specific functions
• The client data can be carried in GFPframes using on of the two adaptationmodes:– Frame-mapped GFP (GFP-F) applicable to
most packet data types– Transparent-mapped GFP (GFP-T) applicable
to 8B/10B coded signals
Connection-Oriented Networks - Harry Perros 81
Frame-mapped GFP
• Variable length frames such as:– Ethernet MAC frames,– PPP/IP packets– HDLC-framed PDUs
can be carried in the GFP payload.• One frame per GFP payload.• Max. size: 65,535 bytes
Connection-Oriented Networks - Harry Perros 82
Transparent-mapped GFP
• Fiber Channel, ESCON, FICON, GigabitEthernet high-speed LANs use 8B/10Bblock-coding to transport client data andcontrol information.
• Rather than transporting data on a frame-by-frame basis, the GFP transparent-mappedmode, transports data as a stream ofcharacters.
Connection-Oriented Networks - Harry Perros 83
• Specifically, the individual characters arede-mapped from their client 8B/10B blockcodes and then mapped into periodic fixed-length GFP frames using 64B/65B blockcoding.
• This reduces the 25% overhead introducedby the 8B/10B block-coding.
• Also, transparent mapping reduces latency,which is important for storage relatedapplications
Connection-Oriented Networks - Harry Perros 84
• The first step, is to decode the 8B/10Bcodes. The 10 bit code is decoded into itsoriginal data or control codeword value.
• The decoded characters are then mappedinto 64B/65B codes. A bit in the 65-bitcode indicates whether the 65-bit blockcontains only data or control characters arealso included
• 8 consecutive 65-bit blocks are groupedtogether into a single superblock.
• A GFP frame contains N such superblocks.
Connection-Oriented Networks - Harry Perros 85
Data over SONET/SDH (DoS)
• The DoS architecture provides an efficientmechanism to transport data coming frominterfaces such as: Ethernet, Fiber Channel,ESCON/FICON over SONET/SDH.
• It relies on a combination of– GFP,– Virtual concatenation, and– Link capacity adjustment scheme (LCAS)
Connection-Oriented Networks - Harry Perros 86
Virtual concatenation• This procedure maps an incoming traffic stream
into a number of individual sub-rate payloads.• The sub-rate payloads are switched through the
SONET/SDH network independently of eachother
• At the destination, they are used to reconstruct theoriginal traffic stream.
Connection-Oriented Networks - Harry Perros 87
Example• Let us consider the case of transporting the
1 GbE signal over SONET/SDH.• According to the specifications, an STS-
48c (2,488 Gbps) has to be used, thusleaving a lot of unused capacity.
• Using the virtual concatenation scheme, 7independent STS-3c (7x155,520 = 1,088)can be employed to carry the 1 GbE signalat full rate.
Connection-Oriented Networks - Harry Perros 88
This works as follows:• At the transmitter the incoming stream is
de-multiplexed and distributed in somefashion over 7 different payloads, each anSTS-3c.
• Intermediate SONET/SDH nodes only seedifferent payloads and they are not awareof the concatenation
• At the destination, the seven flows getmultiplexed into the single original GbEstream.
Connection-Oriented Networks - Harry Perros 89
Link capacity adjustment scheme(LCAS)
• This scheme permits to dynamically adjustthe number of sub-rate payloads allocatedto a traffic stream, whose transmission ratemay vary over time.
• LCAS can be also used when re-routingtraffic due to a failure.